[tlocoh-info] Ancillary variables

maja.bradaric92 at gmail.com maja.bradaric92 at gmail.com
Tue Oct 25 14:50:27 CEST 2016

Hi Andy (and everyone),

Thank you very much for your reply. During past few days I’ve managed to find all of these functions myself and to check how they work on my data. 

However, I encountered some problems. When I used ancillary variable as hull metric in order to make a two dimensional scatter plot and run this function:

hsp2 <- lhs.plot.scatter(Lizard.lhs.k21, x="anv", y="area", anv="Air_temp",col="spiral",bg="black"), 

 I was sent to the debugging environment in R. The message on the top says: “Debug location is approximate because the source is not available”, and when I do the execution, I get my scatter plot. But, when I run the next function:

plot(Lizard.lhs.k21, hpp=T,hsp=hsp2, hpp.classify="hsp"),

 so I can plot my data points and use a scatter plot as a legend, I get the following error message: ”Can’t find values for saved hull scatter plot”. I am guessing I should do something differently in the debug part of the problem?

Also, I think it would be nicer to use mean values of ancillary variables for all data points included in one hull, not only the value of parent point, and I would be really grateful if you or anyone else can help me with that. 



From: Andy Lyons
Sent: 24 October 2016 19:22
To: maja.bradaric92 at gmail.com; tlocoh-info at lists.r-forge.r-project.org
Subject: Re: [tlocoh-info] Ancillary variables

Hi Maja. Good questions. You are correct that there are not many guidelines for using ancillary variables in the documentation, but your instincts on how they could be useful in an analysis seem spot on. As a reminder to others, 'ancillary variable' is the T-LoCoH term for  additional columns or attributes associated with each location. They can be either measured values from a GPS sensor (e.g., temperature), or derived during post-processing from other data (e.g., NDVI values). See lxy.anv.add() and lxy.gridanv.add(). Internally, ancillary variables are saved in columns in the data frame attached to the SpatialPointsDataFrame which contains the locations. All the original points are saved in both locoh-xy and locoh-hullset objects.

One way you can use ancillary variables is for subsetting data, as you suggested. This could be useful for example if you wanted to look at space use patterns for different sets of locations based on an attribute field. To create subsets of hulls, see lhs.filter.anv(), to create subsets of points from a locoh-xy object see lxy.subset() (I can send a code sample if it isn't clear how to use). Subsetting location data could also of course be done prior to turning the locations into a locoh-xy object. 

Ancillary variables can also be used to sort hulls when constructing isopleths. This is one option for differentiating internal space within the 'home range' along a gradient other than density. If your GPS sensor recorded temperature, for example, you could create a utilization distribution that highlights how the individual used the space based on a temperature gradient. IMHO, the ability to create UDs that differentiate space use along behavioral or environmental gradients is one of T-LoCoH's most interesting and underutilized features which has a lot of promise for connecting the concept of a home range to a much broader range of behavioral and ecological questions (e.g., Fieberg and Börger 2012). 

To sort hulls based on an ancillary variable, you would pass sort.metric="anv" to the function lhs.iso.add(), with an additional argument called anv that passes the name of the ancillary variable column name of interest (e.g., anv="temp"). The 'anv' hull metric is defined to be the ancillary variable value of the hull parent point. You may ask why the hull parent point, and not the mean value of all points used to construct the hull, or all points enclosed by the hull? That was just the easiest to implement, but if you were going to sort hulls for isopleths it might be better to define a hull metric for the mean ancillary value for all enclosed points (depends on the pattern you're seeking). Defining new hull metrics is not terribly hard but takes some coding knowledge, contact me for details.
You can also use ancillary variables of the hull parent point as you would any other hull metric - for plotting symbology, in scatter plots, etc. If you see an interesting pattern in a two-dimensional scatterplot of an ancillary variable and another hull metric, you can use that as a legend in a map (e.g., Fig 10 in Lyons et al 2013). 

Hope this helps you think about possible ways for using ancillary variables in an analysis. A lot of this is new territory, but T-LoCoH is all about giving you a flexible set of tools to explore and visualize your data. Let me know if you have any questions or need help with any of the functions.


On 10/19/2016 6:54 AM, maja.bradaric92 at gmail.com wrote:
Hello everyone,
I’ve started to use T-LoCoH just recently and I am still investigating can this package help me answer all the questions I have in my research. I guess this is very basic question, but, there is no so much explained about ancillary variables and what can be done if we have them in our data. 
If I include ancillary variables in my lxy object (temperature, for example), does that mean space use maps will later be created according to the ancillary variable? Or I can use them just to make subsets of the hulls according to certain temperature values? Can I produce behavioural maps according to temperature and possible temperature changes? I don`t really understand how this works, so I would be very grateful for any clarification, in order to be able to continue with my data analysis.
Also, is it possible to include only one ancillary variable, or it can be more of them?
Thanks in advance.

Tlocoh-info mailing list
Tlocoh-info at lists.r-forge.r-project.org

T-LoCoH: A hull-based method for home range construction and spatiotemporal analysis of movement data.
Lyons, A., Turner, W.C., and WM Getz. 2013. Home Range Plus: A Space-Time Characterization of Movement Over Real Landscapes. BMC Movement Ecology 1:2.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/tlocoh-info/attachments/20161025/b3905331/attachment.html>

More information about the Tlocoh-info mailing list